Journal of Autoimmunity 33 (2009) 99–108

Contents lists available at ScienceDirect

Journal of Autoimmunity

journal homepage: www.elsevier.com/locate/jautimm

Gene expression and chromosomal location for susceptibility to Sjo¨gren’s syndrome

Paola Pe´rez a,1, Juan-Manuel Anaya b,1, Sergio Aguilera c, Ulises Urzu´ a a, David Munroe d, Claudio Molina e, Marcela A. Hermoso a, James Michael Cherry d, Cecilia Alliende a, Nancy Olea a, Edward Ruiz-Narva´ez f, Marı´a-Julieta Gonza´lez a,* a University of Chile, Santiago, Chile b Center for Autoimmune Diseases Research (CREA), Corporacio´n para Investigaciones Biolo´gicas, Universidad del Rosario, Medellin, Colombia c INDISA Clinic-Andre´s Bello University, Santiago, Chile d National Cancer Institute, Bethesda, MD, USA e Mayor University, Santiago, Chile f Harvard School of Public Health, Boston, MA, USA article info abstract

Article history: Primary Sjo¨gren’s syndrome (SS) is a chronic inflammatory autoimmune disease affecting mainly the Received 2 April 2009 exocrine glands. Its physio-pathology is poorly understood and most of the knowledge has been related Received in revised form to the inflammatory component. The aim of this work was to evaluate expression profiling in 11 May 2009 fractions enriched in epithelial cells from labial salivary glands (LSGs) of patients with primary SS and Accepted 19 May 2009 identify chromosomal regions harboring susceptibility expressed in epithelial cells. A combined approach of gene expression and genome-wide association study was used. Enriched epithelial cell Keywords: fractions were obtained from LSGs of patients and controls. Amplified total RNA was labeled and Sjo¨gren’s syndrome cDNA microarray hybridized to 10K cDNA microarrays. Results were normalized and subjected to statistical and functional Epithelial cells analysis. A genome-wide microsatellite screen at 10 cM resolution (393 markers) was performed. In Apoptosis salivary gland-epithelial cells from patients 528 genes were expressed differentially in comparison to Interferon controls. Pathways not previously linked to disease were found to be altered. Twenty-eight and 15 genes Genome-wide association study associated with apoptosis were up-regulated and down regulated, respectively. Interferon-related genes, most of which participated in interferon signaling, were also found to be up-regulated. From the genome-wide screen, 6 markers showed evidence of highly significant association with the disease. Of these, five loci harbor genes differentially expressed in patients LSG-epithelial cells. Our results show that in enriched gland-epithelial cells of pSS, both pro-apoptotic/anti-apoptotic and interferon signaling inhibition/stimulation balances may occur. Genes found over-expressed in epithelial cells are candidates for disease susceptibility. Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction metalloproteinases (MMPs), mostly secreted by acinar and ductal cells but not related to the quantity and the proximity of mono- Sjo¨gren’s syndrome (SS) is a chronic autoimmune disease nuclear cells, have been demonstrated in labial salivary glands characterized by lymphocyte infiltration of the exocrine glands, (LSG) from SS patients [4–6]. Based on these findings three phases accompanied by autoantibody production [1]. A key event in the for the development of SS are proposed: first, a preexisting initial process leading to SS seems to be increased epithelial cell condition characterized by genetic susceptibility factors. Second, apoptosis that progresses to subsequent salivary gland lymphocytic a lymphocyte-independent phase which occurs as a consequence infiltration and autoantibody production [2,3]. High levels of matrix of antigens released by necrotic, apoptotic cells or viral trigger, activate innate immune response with cytokine production. Third, a tissue-specific immunological attack by activated T-cells, B-cells * Corresponding author at: Institute of Biomedical Sciences, Faculty of Medicine, and their products against corresponding self-antigens leads to University of Chile, Casilla 70061, Santiago 7, Chile.Tel.: þ56 2 9786017; fax: þ56 2 autoimmune exocrinopathy [1–3]. 737 3158. E-mail address: [email protected] (M.-J. Gonza´lez). The powerful impact of genetic predisposition on susceptibility 1 These authors contributed equally to this work. is usually based on disease concordance rates in monozygotic

0896-8411/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.jaut.2009.05.001 100 P. Pe´rez et al. / Journal of Autoimmunity 33 (2009) 99–108 twins. Although genetic studies in primary SS twins have not been washed with Hanks’ balanced salt solution (Gibco BRL, Carlsbad, CA, performed the observed aggregation of autoimmune diseases in USA) and then centrifuged for 5 min at 1.9 gat4C. Washes were families of patients with primary SS supports a genetic component repeated from 3 to 12 times according to the extent of infiltration in SS etiology [7]. The efforts to unravel the genetic component of observed in each LSG biopsy until no lymphocytes were observed in SS have relied on association studies for disease gene identification. the supernatant or pellet. This was verified by direct microscopic However, robust analyses of candidate gene variants have not been observation of an aliquot of supernatant or pellet of epithelial cells. undertaken and only one linkage study has been reported [8]. In the The short-time, low-speed centrifugation and extensive washes absence of chromosomal regions identified by linkage studies, resulted in a pellet containing mostly epithelial (acini and ducts) research has focused on candidate gene approaches (by biological material. The final pellet was split to extract RNA for gene expression plausibility) rather than on positional approaches. To date, there is assays (see Section 2.3) and for morphological analysis as follows: not any specific gene for primary SS, and all of the polymorphisms aliquots were taken at each isolation step and fixed at 4 C for 18 h in associated with primary SS have been also associated with other a glutaraldehyde 2.5%-sodium caccodylate 0.1 M pH 7.4 buffer. Then, autoimmune diseases [9]. they were rinsed in a caccodylate buffer and post-fixed in a 1% DNA microarray technology allows gene expression information osmium tetroxide-0.1 M caccodylate buffer pH 7.4 for 1 h. Dehy- on a wide-genome basis to be obtained. A few reports have dration in graded series of ethanol was done before infiltration in an described transcriptional profiles of whole LSG from SS patients epoxy resin. Acinar fraction sections (1 mm) were obtained by means [10–12]. A neglected aspect of these studies is that the amount of of a Porter-Blum ultramicrotome (Sorvall MT-2, Thermo Electron mononuclear infiltrate in LSG varies between SS patients [13]. Corporation, Asheville, NC, USA) and stained with toluidine blue in Therefore, these gene expression profiles actually represent 1% sodium borate. In addition, the quality of the resultant cellular variable combinations of different cell types. To overcome the fractions was verified by reverse transcriptase-polymerase chain interference of infiltrated cells on epithelial cell gene expression, reaction (RT-PCR) [5] of marker genes such as: MS4A1, CD3G, ACTA2, we developed a method to obtain fractions enriched in epithelial and KRT18 as markers for B-lymphocytes, T-lymphocytes, myoepi- cells from LSGs and to determine their gene expression profiles thelial cells and acinar-ductal cells, respectively. Primers and using DNA microarrays. Through a genome-wide association study reaction conditions of each gene are described in Supplementary (GWAS) we next sought to identify chromosomal regions of SS data STable 2. susceptibility harboring expressed genes in LSG of SS patients. 2.3. RNA isolation, amplification and fluorescent cDNA labeling 2. Methods Total RNA was procured with the RNeasy Mini Kit (Qiagen, 2.1. Patients Valencia, CA, USA). Quality was verified with the Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA). RNAs were All participating subjects were informed about the aims and amplified with the RiboAmp amplification kit (Arcturus, Sunnyvale, procedures of the present study and then signed a written consent CA, USA) following the supplier’s protocol. 2 mg of amplified RNA approved by the Ethics Committee of the School of Medicine, were labeled by reverse transcription in a 45 mL reaction containing University of Chile and the ‘‘Corporacio´ n para Investigaciones 1 first strand buffer, 2 mg random primers, 2 mg oligo-dT, 0.5 mM Biolo´ gicas’’ (CIB), in Medellı´n, Colombia. All the patients met the of dATP, dGTP and dTCP, 0.2 mM dTTP, 0.08 mM of either Cy3 or Cy5 American-European diagnosis criteria for primary SS [14]. of dUTP, 0.01 mM dithiothreitol, and 500 U of SuperScript II enzyme For the microarray study, patients (n ¼ 9) were classified (Invitrogen, Carlsbad, CA, USA). After incubation at 42 C for 1 h, the according to the focus score (FS) [13]: FS 1–2 (n ¼ 2, age range reaction was halted with EDTA. Residual RNA was hydrolyzed with 43–60 years old, mean 51); FS 3–4 (n ¼ 3, age range 35–66 years NaOH at 65 C for 20 min. NaOH excess was neutralized by addition old, mean 48); FS higher than 4 (n ¼ 4, age range 35–64 years old, of Tris–HCl 1 M, pH 7.5 buffer. The labeled cDNA was purified by mean 54). Control subjects (n ¼ 6, age range 23–43 years old, mean ethanol precipitation in the presence of 20 mg of glycogen. Absor- 35) were selected from individuals not meeting the SS classification bance readings at 260, 550 and 650 nm were measured to estimate criteria. All control subjects were negative for rheumatoid factor, labeling efficiency. antinuclear antibodies and anti-Ro and anti-La antibodies. Mild and nonspecific inflammation (Chisholm and Mason grade 1) was 2.4. Microarray manufacture and hybridization observed in lip biopsies from the control group. For the GWAS, 85 women with primary SS and 178 ethnic-, Microarrays were printed on poly-lysine coated slides using gender-, age- and socioeconomic-matched controls, belonging to a MicroGrid II arrayer (Biorobotics, Boston, MA, USA) at the a homogeneous population [15,16], were included (Supplementary Laboratory of Molecular Technology, SAIC-NCI Frederick. Spotted data STable 1). DNAs consisted of 9961 distinct PCR-amplified human cDNAs prepared from the IMAGE-EST clone collection. The printed pattern 2.2. Preparation of a cell fraction enriched in epithelial cells and the complete clone list are available at the NCI’s Microarray from labial salivary glands Database (‘‘mAdb’’) (http://nciarray.nci.nih.gov/cgi-bin/ Hsatcincyteform.cgi). Labeled test and reference (a pool of LSGs were digested in 10 mL of DMEM pH 7.4 (Gibco BRL, Carls- control individuals) cDNAs were combined at approximately Cy3/ bad, CA) containing 1 mg/mL of collagenase IV (Sigma–Aldrich, St. Cy5 equimolar quantities and co-hybridized in 38 ml solution Louis, MO) for 30 min at 37 C and shaken at 120 cycles/minute. containing final concentrations of 1 mg/ml human Cot-1 DNA, 1 mg/ Then, the suspension was incubated in fresh collagenase-containing ml poly dA, 0.1% SDS and 3.5 SSC. This mixture was denatured at medium for 45 min. Mechanical disaggregation was done by passing 99 C for 3 min, cooled at room temperature, and deposited onto the cell suspension through a series of three Pasteur pipettes of the array surface. Microarrays were placed in hybridization decreasing tip diameters (2.0, 1.0 and 0.5 mm). After the third chambers and incubated at 65 C for 14–18 h in a water bath. pipetting round, the suspension was centrifuged for 5 min at 1.9 g Washes in 2 SSC-0.1% SDS, 1 SSC, 0.2 SSC and 0.05 SSC at 4 C, the supernatant discarded and the pellet re-suspended in were performed, 1 min each. Microarrays were dried and a collagenase-free medium. Epithelial cell suspensions were further scanned at 10 mmresolutioninaGenePix4000Bscanner P. Pe´rez et al. / Journal of Autoimmunity 33 (2009) 99–108 101

(Axon Instruments, Union City, CA). Voltage was adjusted to 2.7. DNA preparation and genotyping obtain maximal signal intensities with minimal (<0.1%) probe saturation. Pseudocolor microarray images were saved in tiff format. Genomic DNA was extracted from 10 mL of an EDTA-anti- coagulated blood sample using the classical salting out protocol. 2.5. Validation of microarray results Three hundred ninety three polymorphic Marshfield microsatellite markers with an average spacing of w10 cM from Marshfield Quantitative PCR (Q-PCR) assays of selected genes, as well as Screening Set 16 were genotyped by the Marshfield Center for immunohistochemical localization and expression of the MSN Medical Genetics (http://research.marshfieldclinic.org/genetics/). were performed as microarray data validation. Briefly, specific primer pairs for the 30-mRNA ends of test (HLA-DPA1, 2.8. Genetic analysis MMP-9, MSN and VIM) and reference (GAPDH) mRNAs were obtained with the Assay-by-design software (Applied Biosystems, Allele frequencies in each microsatellite marker locus in the SS Foster City, CA, USA). The cDNAs were prepared by reverse tran- cases and matched controls were estimated by the gene-counting scribing 1 mg of amplified RNA with the Superscript II enzyme method. We assessed whether there was any hidden population (Invitrogen, San Diego, CA, USA). Q-PCR assays were conducted in stratification by using the model-based clustering method under an 384-well microtiter plates in 10 mL final volumes. PCR cycles in the admixture model as implemented in the Structure software version Applied Biosystems ABI-7900 SDS thermocycler were 50 C for 2.2 [19]. Briefly, the admixture model assumes that different K 2 min, 95 C for 10 min and 40 cycles of 95 C for 10 s and 60 C populations contribute to the genomic composition of our sample for 1 min. Quantification was based on CT values, which represent [19]. Each individual in the sample had different genomic propor- the PCR cycle at which an increase in reporter fluorescence above tions of each of the K populations. We ran the admixture model baseline signal can be detected. Normalization employed the under three different number of populations (K ¼ 1, 2, and 3) using GAPDH gene value as reference transcript assayed under identical 10,000 Markov Chain Monte Carlo simulations for each K. The conditions respective to the gene of interest. Numerical conver- structured population association test (STRAT) [20], using 10,000 sion of CT values to microarray ratios was based in reverse DDCT- Markov Chain Monte Carlo simulations for each marker, was Sample values, were DDCT-Sample ¼ DDCT-Sample DCT-Reference. For applied to test for differences in the microsatellite allele frequencies MSN protein validation, samples were fixed in 1% para- between SS cases and controls. We used a –log(p-value) ¼ 2, formaldehyde for 6 h and then embedded in paraffin. Sections equivalent to a p-value ¼ 0.01, to determine significance. were assayed for epitope recovery by heating at 95 C for 20 min in 10 mM citrate buffer (pH 6.0). Endogenous peroxidase activity 3. Results was quenched with 0.3% H2O2 in methanol, and nonspecific binding was blocked with 0.25% casein during 30 min. Sections 3.1. Analysis of fractions enriched in acinar and ductal were incubated with a monoclonal antibody against human MSN epithelial cells (BD Biosciences, San Jose, CA) followed by incubation with a rabbit anti-mouse Alexa fluor 488 antibody (Molecular Probes, In order to examine the molecular events occurring in the LSG- Invitrogen, CA). For nuclear counterstaining, the sections were epithelial cells of SS patients, we obtained a cellular fraction incubated in 1 mg/mL propidium iodide, 1 mg/mL RNAse A in enriched in acinar and ductal cells. The quality of the fraction was phosphate buffer pH 7.4 for 1 min. The pre-immune controls were analyzed morphologically. Fig. 1 shows panoramic views of sections also performed. of enriched epithelial fractions from one control (A) and two patients (C and D). The control sample displayed many acini with 2.6. Microarray data extraction and analysis narrow lumen, and acinar cells with typical polarized morphology showing the nucleus located at the basal region and the secretory All microarray data presented here is in the two-color compar- granules at the apical region. Details of a seromucous acinus are ative format, i.e., it corresponds to the log2-expression ratio of the shown at higher magnification in Fig. 1B. In SS patients, acini fluorescent signal in the test sample over the fluorescent signal in clusters of variable-size and characteristic luminal dilation were the reference sample for every single gene spotted on the array. observed in association with intercalated ducts; the inset shows Microarray tiff images were extracted as GenePix result (gpr) files details of a seromucous acinus (Fig. 1C). Fig. 1D shows a striated using the GenePix Pro III software (Molecular Devices CA). Gpr files duct among several acini (see details of striated duct in the corre- and jpeg array images were deposited at mAdb. Data were subse- sponding inset). Only in one from 9 SS patients a discrete cluster of quently subjected to print-tip (local) Loess normalization and scale inflammatory cells was found. To confirm our morphological correction with the DNMAD tool. (http://dnmad.bioinfo.cnio.es) observations markers for each cell type were monitored by RT-PCR [17]. Genes which were present in less than 85% of the arrays were (see Methods) which are shown in Fig. 1E (control) and 1F (SS excluded from the analysis. Missing values were imputed using patient). Lymphocytic cell markers (MS4A1 and CD3G) were KNN-impute (weighted K-nearest neighbors) at GEPAS (http:// undetectable in the enriched epithelial cell fraction (lanes E) of gepas.bioinfo.cnio.es/cgi-bin/preprocess) [18]. At this point, the control samples (Fig. 1E), whereas in an SS patient, the dataset size was reduced to 4866 clones. Patient data was correspondent bands (lanes E) were weaker than in the whole compared with control data using a limma t-test with adjustment salivary gland (lanes G, Fig. 1F). On the contrary, epithelial markers for multiple testing performed with the tool Pomelo II (available at: (KRT18 and ACTA2) showed higher levels indicating epithelial cell http://pomelo2.bioinfo.cnio.es/). A total of 627 cDNA clones were enrichment in both groups. detected as differentially expressed using an adjusted p < 0.05 as cut-off. After excluding repeats, transcribed loci and unknowns, the 3.2. Genome-wide transcriptional profile of fractions enriched 627 list was reduced to 528 known genes that were the subject of in LSG-epithelial cells from SS patients functional analysis. Correlation of transcriptional profiles with lymphocytic infiltrate (i.e., FS) was done with the multiple testing For all hybridizations, a pool of total RNA from control individ- tools Pomelo (http://pomelo.bioinfo.cnio.es/) with false discovery uals was used as the reference, so the differentially expressed genes rate (FDR) control. in SS patient samples were obtained directly. Repeated dye-swap 102 P. Pe´rez et al. / Journal of Autoimmunity 33 (2009) 99–108

Fig. 1. Morphological and molecular analysis of enriched LSG-epithelial cell fractions. Images A–D show LSG sections stained with toluidine blue of one control subject (A, B) and two SS patients (C, D). From each preparation, randomly chosen fields are shown. Arrowhead in A indicates a seromucous acinus shown in B. Arrowheads in C and D indicate the corresponding insets. s: serous acinar region, m: mucous acinar region, L: acinar lumen, ID: intercalated duct and SD: striated duct. Bars: 21 mm. Panels E and F show RT-PCR amplification products of cell-type marker genes (CD3G, MS4A1, KRT18, and ACTA2) and the endogenous control gene (GAPDH) from whole LSG (lanes G) and enriched LSG- epithelial cell fractions (lanes E) obtained from representative controls (panel E) and SS patients (panel F). St: 100 bp ladder.

(i.e., 4 replicas for each sample) was made, resulted in 52 micro- however 15 anti-apoptosis were also up-regulated. The genes BCL2, array data files which were deposited in mAdb. The mean BIRC2, BIRC3, CASP10, CFLAR, CSF2RB, FADD, IL1R1, NFKB1, NFKBIA, percentage of ‘‘found’’ spots was 87 13% and the mean signal-to- and TNFRSF6 belong to the KEGG’s Apoptosis pathway. background ratios were 5.93 and 7.49 in the Cy3 and Cy5 channels Additionally, a group of 11 genes related to interferon (IFN) respectively. Prior to statistical analysis, data were normalized activity (ICSBP1, MX1, IFITM1, IFIT2, IFI16, IRF7, NFKBIA, IRF9, IFNGR1, using Loess analysis [17]. Normalized data were pre-processed and CEBPG, DAP) were found over-expressed in SS samples (Table 3). imputed resulting in a subset of 4866 genes. Differentially expressed genes were extracted with a limma t-test under control 3.3. Validation of microarray results on selected genes of multiple test, a typical issue in microarray data analysis. This analysis resulted in 528 known genes and 74 transcribed loci and Our recent work has demonstrated alterations in both epithelial unknown clones. Eighty-one percent of the total list was up-regu- cells and basal lamina of LSG from SS patients [4–6]. The genes lated and most of them are considered typical epithelial genes. In MSN, VIM, ARPC2, KRT14, CLDN10, CLDN4, MMP9 and VIL2 were this list appeared genes or their previously involved in SS present in the gene list and have a role in maintenance and (i.e., SELL (CD62L), VCAM1, ICAM1, CCL4, CX3CL1, LTF and IL6) [9]. organization of cell architecture. Accordingly, earlier reports of our Table 1 shows the mean ratio values of the 20 top ranked genes. group showed that MMP9 over-expression and higher levels of Ten of these 20 have been previously associated with SS active MMP9 protein co-occur in severe SS patients, which yet [5,9,11,21–27], whereas ten (IGHG1, RARRES1, NCF1, SYT1, HLA-DMA, conserve high amount of parenchyma [4,5]. In addition, MHC-class PCOLN3, COL3A1, PRG1, COTL1, BF) are described here for the first II genes have been recurrently involved in this disease [9]. There- time associated to primary SS. This new gene set included all top fore, HLA-DPA1, MMP-9, MSN and VIM genes were selected to ranked down-regulated genes. A complete list of all functions perform a side-by-side comparison of expression levels measured obtained from , ‘‘Kyoto Encyclopedia of Genes and by microarray versus real-time PCR assays in two SS patients. As it Genomes’’ (KEGG) and Biocarta Pathway databases is shown in is shown in Fig. 2A, Q-PCR and microarray results showed to be very Supplementary data STable 3. consistent in terms of magnitude and direction. The only exception Apoptotic genes represent over 10% of all known 528 differen- was the MMP9 level in P3, which change in magnitude but no in tially expressed genes. Table 2 show the relative expression of pro direction. We also examine MSN levels in subjects C1, P2, P4 and P6 and anti-apoptosis genes with significative up- and down-regula- employing 3 dilution of template cDNA (Fig. 2B). In agreement to tion. Up-regulation of 28 pro-apoptotic genes was predominant; microarray data, MSN was found over-expressed in all studied P. Pe´rez et al. / Journal of Autoimmunity 33 (2009) 99–108 103

Table 1 Top ranked expressed genes in enriched LSG-epithelial cell fractions from SS patients.

Gene symbol, description Adjusted p-valuea Gene expression (log2) Chromosomal Location Previously studied in SS IGHG1, Immunoglobulin heavy constant gamma 1 (G1m marker) 0.0001 2.65 14q32.33 No HLA-DRA, Major histocompatibility complex, class II, DR alpha 0.0005 2.20 6p21.3 Yes [9,11] RARRES1, Retinoic acid receptor responder (tazarotene induced) 1 0.0070 2.19 3q25.32-q25.33 No HLA-A, Major histocompatibility complex. class I A 0.0000 2.05 6p21.3 Yes [9,11,29] NCF1, 47-kD autosomal chronic granulomatous disease protein 0.0000 1.97 7q11.23 No SYT1, Synaptotagmin I 0.0014 1.82 12cen-q21 No UBD, Ubiquitin 0.0007 1.74 6p21.3 Yes [11] CXCL9, Mig Humig chemokine targeting T-cells 0.0070 1.68 4q21.1 Yes [21,22] HLA-DRB5, Major histocompatibility complex. class II. DR beta 4 0.0016 1.54 6p21.3 Yes [27] HLA-DRB1, Major histocompatibility complex. class II. DR beta 4 0.0004 1.49 6p21.3 Yes [9,27,31,32] MMP9, Matrix metalloproteinase 9 gelatinase B 0.0007 1.48 20q11.2-q13.1 Yes [5] HLA-DPA1, Major histocompatibility complex. class II. DP alpha 1 0.0004 1.47 6p21.3 Yes [11] HLA-DMA, MHC-Class II DM alpha 0.0074 1.43 6p21.3 No CD69, CD69 early activation antigen 0.0038 1.24 12p13-p12 Yes [26] LTF, Lactotransferrin 0.0037 1.21 3q21-q23 Yes [23,24] PCOLN3, Procollagen type III N-endopeptidase 0.0015 1.02 16q24.3 No COL3A1, collagen type III alpha 1 0.0007 0.91 2q31 No PRG1, Proteoglycan 1. secretory granule 0.0007 0.86 19q13.2 No COTL1, Coactosin-like 1 Dictyostelium 0.0008 0.54 16q24.1 No BF, B-factor. Properdin 0.0039 0.33 6p21.3 No

a The shift represents the log2 difference between the mean ratio values of all replicates of the highest (4) minus the lowest (0) focus score patient group, for the indicated gene.

samples. Finally, the MSN protein level was assayed by both pro-apoptosis and 15 anti-apoptosis related genes up-regulated immunohistochemistry and Western-blot. Green fluorescence for (Table 2). CASP10 up-regulation suggests that the SS-apoptotic MSN was located at apical pole of epithelial cells (acini or ducts) in pathway could be mediated by death receptors but modulated by both salivary glands from SS patients or control individuals the action of CFLAR, MCL1 and BIRC3 [36]. CFLAR encodes FLICE- (Fig. 2C). The Western-blot pattern for MSN showed two bands at inhibitory protein (FLIP), an inhibitor apoptotic protein (IAP), which 80 and 78 kDa, these bands correspond to VIL2 and MSN respec- inhibits caspase-10, the enzyme that is the first member of this tively, which are comparable with the pattern showed by positive pathway. MCL1 belongs to the BCL2 family and down-regulates control (Fig. 2D). The relative levels of MSN are 1.9 higher in receptor-mediated apoptosis [34].BIRC3 encodes for IAPC-2, which patients than controls. Both proteins are part from the family selectively inhibits caspases 3 and 7 and also activates their ubiq- proteins ERM, and its identity sequence is about 78%. uitination [36]. CUL5 encoded a scaffold protein and a member of the E3 ubiquitin-protein ligase complex (Skp1-Cullin-F-box-typeE3 3.4. Genome scan identifies potential SS susceptibility loci ubiquitin-protein ligase). This complex recognizes phosphorylated harboring LSG-epithelial cell expressed genes I-kappa-B and promotes its degradation, thus assisting in the activation of NFKB [37,38]. In this regard, PKRC phosphorylates The log (p-value) of the microsatellite markers on the different RelA, part of the active NFKB complex, by triggering its transcrip- human is shown in Fig. 3. Six out of the 393 markers tional activity [39]. Finally, CUL5 in combination with NFKB and showed evidence of highly significant associations with primary SS IAPC-2 participates in processes that support cell survival. in the population studied (Table 4). As expected, there was no The intrinsic apoptotic pathway is activated by diverse stimuli evidence of hidden population stratification in the cohort studied. such as oxidative stress, cytokines, nitric oxide, the inflammatory It is clear that almost 100% of the gene copies in our sample come infiltrate and also as a response to the anchorage loss of acinar cells from a single population (Supplementary data STables 4 and 5). from its basal lamina. The product of the BCL2A1 gene, located in the mitochondrial membrane, is a constituent of this pathway and 4. Discussion has a cyto-protector effect. BNIP3L was also up-regulated. The product of this gene forms a complex with Bcl-2, which has a pro- This is the first report on the use of an enriched LSG-epithelial apoptotic effect that induces the release of cytochrome-c and cell fraction in order to explore gene expression signatures from proteins blocking IAPs activity (i.e. DIABLO) [40,41]. A recent study patients with primary SS and to identify SS susceptibility loci by using the salivary gland-epithelial cell line A-253 showed that GWAS. Therefore, our results constitute an important source of auto-antibodies from SS patients induce both the intrinsic and the information which will help to direct future research on SS extrinsic apoptotic pathways [42]. In human submandibular gland pathogenesis. Cell separation techniques from tissues containing cells, TNF-a and IFN-g trigger the activation of the mitochondrial multiple cell types are usually unable to accomplish absolute death inducer, tBid, and of the intrinsic apoptotic pathway [43].We purity. The method described here produced higher yields of acini propose that the combined action of auto-antibodies and cytokines and ducts from LSG and a very low number of lymphocytic cells. might modulate apoptosis induction in epithelial cells from salivary Molina et al. showed in SS patients the invasion of TCD4þ and glands of SS patients. To our knowledge, neither caspase-10 nor the TCD8þ lymphocytes in ducts whose basal lamina were disrupted balance of pro- and anti-apoptotic molecules participating in both (35). This finding may explain CD 69 expression in the present apoptotic pathways has been studied in SS. study. Recent reports from our laboratory have demonstrated alter- Apoptosis has been postulated as a mechanism for gland ations in both LSG-epithelial cells and basal lamina from SS patients damage in SS and alteration of secretory function [28,30]. However, [6,35,44]. The genes MSN, VIM, ARPC2, KRT14, CLDN4, MMP9 and the involved mechanisms are not completely understood and, in VIL2 were present in the list of 528 genes and have a role in cell some cases, they are contradictory. In SS patients, we found 28 architecture, maintenance and organization [45–47]. MSN, VIL2 and 104 P. Pe´rez et al. / Journal of Autoimmunity 33 (2009) 99–108

Table 2 Relative expression of apoptosis-related genes in enriched LSG-epithelial cell fractions from SS patients.a

Gene symbol, description Adjusted p-value Gene expression (log2) Chromosomal Location Pro-apoptosis Up-regulated MMP9, Matrix metallopeptidase 9 0.0007 1.67 20q13.12 SRGN, Serglycin 0.0008 1.35 10q21.3 BTG1, B-cell translocation gene 1, anti-proliferative 0.0000 1.30 12q21.33 CUL3, Cullin 3 0.0008 1.23 2q36.2 PMAIP1, phorbol-12-myristate-13-acetate-induced protein 1 0.0005 1.17 18q21.32 MX1, myxovirus (influenza virus) resistance 1, interferon-inducible protein p78 0.0013 1.15 21q22.3 IFI16, interferon, gamma-inducible protein 16 0.0011 0.99 1q22 CDKN1B, cyclin-dependent kinase inhibitor 1B (p27, Kip1) 0.0000 0.90 12p13.1 TIMP3, TIMP metallopeptidase inhibitor 3 0.0086 0.90 22q12.3 ZC3H12A, Zinc finger CCCH-type containing 12A 0.0059 0.89 1p34.3 BNIP3L, BCL2/adenovirus E1B 19 kDa interacting protein 3-like 0.0055 0.83 8p21 CUL5, Cullin 5 0.0039 0.80 11q22.3 NFKBIA, nuclear factor of kappa light polypeptide gene enhancer in B-cells inhibitor, alpha 0.0109 0.80 14q13.2 EIF2AK2, eukaryotic translation initiation factor 2-alpha kinase 2 0.0011 0.75 2p22.2 GADD45G, growth arrest and DNA-damage-inducible, gamma 0.0221 0.69 9q22.2 CASP10, caspase 10, apoptosis-related cysteine peptidase 0.0253 0.65 2q33-q34 NFKB1, nuclear factor of kappa light polypeptide gene enhancer in B-cells 1 (p105) 0.0416 0.64 4q24 STK4, Serine/threonine kinase 4 0.0015 0.62 20q13.12 EP300, E1A binding protein p300 0.0058 0.52 22q13.2 CUL1, Cullin 1 0.0072 0.52 7q36.1 NAE1, NEDD8 activating enzyme E1 subunit 1 0.0258 0.48 16q22.1 FADD, Fas (TNFRSF6)-associated via death domain 0.0350 0.45 11q13.3 RRAGA, Ras-related GTP binding A 0.0273 0.44 9p22.1 PPP2CA, protein phosphatase 2 (formerly 2A), catalytic subunit, alpha isoform 0.0418 0.44 5q31.1 ATG5, autophagy related 5 homolog (S. cerevisiae) 0.0240 0.43 6q21 SMAD3, SMAD family member 3 0.0474 0.43 15q22.33 CTNNBL1, Catenin, beta like 1 0.0296 0.42 20q11.23 TIAL1, TIA1 cytotoxic granule-associated RNA binding protein-like 1 0.0343 0.40 10q

Down regulated PDIA3, Protein disulfide isomerase family A, member 3 0.0364 0.54 15q15.3 PDCD4, Programmed cell death 4 0.0362 0.61 10q25.2 DAP, death-associated protein 0.0039 0.87 5p15.2

Anti-apoptosis Up-regulated ANXA4, Annexin A4 0.0005 1.46 2p14 IL6, interleukin 6 (interferon, beta 2) 0.0016 1.28 7p21 TNFAIP3, tumor necrosis factor, alpha-induced protein 3 0.0032 1.24 6q23.3 BIRC3, baculoviral IAP repeat-containing 3 0.0191 1.21 11q22.2 SOD2, superoxide dismutase 2, mitochondrial 0.0074 0.98 6q25.3 BCL2A1, BCL2-related protein A1 0.0100 0.80 15q25.1 ACTN3, actinin, alpha 3 0.0187 0.68 11q13.1 IGF1R, insulin-like growth factor 1 receptor 0.0246 0.62 15q26.3 SEMA4D, sema domain, immunoglobulin domain (Ig), transmembrane domain 0.0106 0.58 9q22.2 (TM) and short cytoplasmic domain, (semaphorin) 4D CDC2, Cell division cycle 2, G1 to S and G2 to M 0.0444 0.55 10q21.2 SOD1, superoxide dismutase 1, soluble 0.0063 0.52 21q22.11 TAX1BP1, Tax1 (human T-cell leukemia virus type I) binding protein 1 0.0416 0.50 7p15.2 BCL2, B-cell CLL/lymphoma 2 0.0061 0.49 18q21.3 BIRC2, baculoviral IAP repeat-containing 2 0.0410 0.47 11q22 SON, SON DNA binding protein 0.0331 0.45 21q22.11

Down regulated BCL2L2, BCL2-like 2 0.0311 0.44 14q11.2 DHCR24, 24-dehydrocholesterol reductase 0.0074 0.53 1p32.3 PHB, prohibitin 0.0124 0.63 17q21.33 NME1, non-metastatic cells 1, protein (NM23A) 0.0014 0.64 17q21.33 KRT18, keratin 18 0.0480 0.65 12q13.13 PHLDA1, pleckstrin homology-like domain, family A, member 1 0.0074 0.81 12q21.2

Dual apoptotic role CLU, Clusterin 0.0004 1.07 8p21.1 FAS, TNFRSF6 (TNF receptor superfamily, member 6) 0.0258 0.75 10q24.1 CFLAR, CASP8 and FADD-like apoptosis regulator 0.0403 0.65 2q33.1 BARD1, BRCA1 associated RING domain 1 0.0356 0.48 2q35

a A t-test was performed between patient and control samples with the tool Pomelo (available at http://pomelo2.bioinfo.cnio.es/). The adjusted p-value shown is the output of a false discovery rate (FDR) approach aimed to control multiple testing issues. The gene expression value corresponds to the difference between gene expression averages of patients minus control subjects in the log2 scale. Table 3 Top ranked, interferon-signaling associated genes in enriched LSG-epithelial cell fractions from SS patients.

Gene symbol, description Adjusted p-value Mean Gene Chromosomal Previously a expression (log2) Location studied in SS ICSBP1, ICSBP Interferon consensus sequence binding protein, IRF8 0.017 1.47 16q24.1 Yes [11] MX1, myxovirus (influenza virus) resistance 1, interferon-inducible protein p78 0.001 1.15 21q22.3 Yes [34] IFITM1, interferon induced transmembrane protein 1 (9–27) 0.001 1.13 11p15.5 Yes [2,3,33] IFIT2, Interferon-induced protein with tetratricopeptide repeats 2 0.009 1.11 10q23-q25 Yes [33] IFI16, Interferon, gamma-inducible protein 16 0.001 0.99 1q22 Yes [4,5,33] IRF7, Interferon regulatory factor 7 0.004 0.93 11p15.5 Yes [6,7,33] NFKBIA, IkB alpha 0.011 0.80 14q13 Yes [8] IRF9, interferon regulatory factor 9, ISGF3G 0.015 0.74 14q11.2 Yes [11] IFNGR1, interferon gamma receptor 1, CD119 0.022 0.61 6q23.3 Yes [11] CEBPG, CCAAT/enhancer binding protein (C/EBP), gamma 0.041 0.45 19q13.11 No DAP, death associated protein 0.004 0.87 5p15.2 No

a A t-test was performed between patient and control samples with the tool Pomelo (available at http://pomelo2.bioinfo.cnio.es/). The adjusted p-value shown is the output of a false discovery rate approach aimed to control multiple testing issues. The gene expression value corresponds to the difference between gene expression averages of patients minus control subjects in the log2 scale.

5 5 A P4 Array P6 Array Q-PCR Q-PCR 4 4

3 3 ratio ratio 2 2 2 2 Log Log 1 1

0 0 HLA-DPA1 MMP9 MSN VIM HLA-DPA1 MMP9 MSN VIM

B Control Patients 1.6 C1 P4 P2 P6 Moesin 1.2 394bp 0.8

GAPDH 0.4 600 bp (Moesin/GAPDH) 0.0 Relative expression cDNA concentration C1 P4 P2 P6

C Control (C2) Patient (P4)

D Controls Patients kDa C+ C1 C2 P3 P4 P5 P6

78 moesin

47 β-actin

Fig. 2. Validation of the microarray results on selected genes by mRNA and protein expression. The genes HLA-DPA1, MMP9, MSN and VIM were among the statistically significant up-regulated transcripts and further validated as follows: in A), the microarray and the Q-PCR (RealTime PCR) results are contrasted for the above-mentioned genes in patients P4 and P6. In B), the RT-PCR amplification product of the moesin mRNA is shown for control (C1) and patients (P4, P2 and P6). The 3 lanes for each subject correspond to decreasing cDNA concentrations employed in the assays, as described. The reaction product of GAPDH is also shown for all assays. Bar graphic shows the densitometric analysis of semi- quantitative RT-PCR products expressed as mean MSN/GAPDH ratios of 3 independent assays. Whiskers above the bars indicate the standard error. In C), immunofluorescence images from LSG sections showing the MSN distribution in the acinar lumen for control individual (C2) and SS patient (P4). Nuclei are stained with propidium iodide (Bars ¼ 15 mm). In D), protein levels of MSN measured by Western blot are shown for controls subjects and SS patients. The bands of 78 and 47 kD correspond to MSN and b-actin, respectively, the latter used as gel loading control. 106 P. Pe´rez et al. / Journal of Autoimmunity 33 (2009) 99–108

Fig. 3. Distribution of the –log(p-values) of the microsatellite markers on the different human chromosomes. Differences in the microsatellite allele frequencies were determined using 10,000 Markov Chain Monte Carlo simulations. We used a threshold of –log(p-value) ¼ 2.0 (p-value ¼ 0.01), which is indicated by the horizontal line.

CLN4 are located at the apical pole of epithelial cells. Thus their genes containing the ISRE sequence (IFN sequence regulatory up-regulation could probably explain the disruption of this cellular element) such as: IRF7, G1P2, GBP1 and IFITM1 [48]. IRF-7 partici- compartment in SS. Real-time PCR data confirmed microarray pates in the expression of IFN-activated genes, cytokines and some results for MMP9 (Fig. 2A). MMP9 over-expression has been chemokines [49] and it has been demonstrated that plays an described previously in SS LSG in relationship to the structural essential role for a positive-loop feedback in IFN signaling [48,49]. disorganization of acini and ducts [4]. Related studies have demonstrated that auto-antibodies (anti-SSA A group of 11 genes related to IFN activity (ICSBP1, MX1, IFITM1, and/or SSB) occurring in the serum of SS patients can induce IFN IFIT2, IFI16, IRF7, NFKBIA, IRF9, IFNGR1, CEBPG, DAP) were found expression in plasmacytoid dendritic cells (pDCs) [41,50]. LSG- up-regulated in primary SS samples (Table 3). Most of them ductal cells from SS patients produce high levels of CXCL9 (Mig) and participate in IFN signaling which suggests over-activation of this CXCL10 (IP-10) chemokines while CXCL12 (SDF-1) is expressed pathway [33,48]. IFITM1 is related to the over-expression of the constitutively in these glands and CXCL9 is highly expressed [22]. P48-subunit (ISGF3G) from the ISGF3 complex. In response to IFN CXCL12 is an excellent chemoattractant for pDCs and acts in receptor stimulation, ISGF3 (STAT2) activates the expression of synergy with CXCL9 and CXCL10 [51]. These data could explain in part the periductal location of the focus in SS patients, in addition they suggest a close cross-talk between inhibitory and stimulatory Table 4 IFN signaling to form and maintain periductal foci. Microsatellite markers with significant differences in allele frequencies in SS cases Genes related to IFN signaling have been reported in the past in and controls, and candidate’s genes expressed in enriched LSG-epithelial cell frac- SS [11,12,33,50]. This observation led to the proposal that a persis- tions from SS patients. tent IFN signal would induce a vicious cycle allowing a permanent region Marker p-valuea Candidate Gen (s)b activation of toll-like receptors as a response to viral RNA [12]. 1p34.2 D1S3721 0.006 LAPTM5, ZC3H12A, NASP Recently, the activation of toll-like receptors has also been related 3q13.31-q13.32 D3S2460 0.007 LTF with cleaved proteins from extracellular matrix from inflamed 7p22.2-p22.1 D7S1819 0.008 IL6 tissue, initiating an inflammatory response even in the absence of 8q11.23 D8S1110 0.004 pathogens and infiltrating immune cells [52]. Interestingly, 10p13 D10S1430 0.0004 CD44 14q12 D14S608 0.001 BCL2L2, NFKBIA, IRF9 previous studies in our laboratory have showed an increase of these cleaved proteins in salivary glands of SS patients [6]. a Differences in the microsatellite allele frequencies were determined using 10,000 Markov Chain Monte Carlo simulations. We used a combined approach of gene expression and GWAS to b As described in Santa Cruz Project Working Draft (http:// identify SS susceptibility regions harboring differentially expressed genome.ucsc.edu) and The National Center for Biotechnology Information Map genes. From the genome-wide screen, 6 markers showed evidence Viewer (www.ncbi.nlm.nih.gov/projects/mapview) LAPTM5: lysosomal multi- of highly significant associations with SS. It is very unlikely that our spanning membrane protein 5, ZC3H12A: zinc finger CCCH-type containing 12A, results are confounded by a hidden population structure because NASP: nuclear autoantigenic sperm protein (histone-binding), LTF: lactotransferrin, IL6: interleukin-6, BCL2L2: BCL2-like 2, NFKBIA: nuclear factor of kappa light poly- there was no evidence of population stratification in the data. peptide gene enhancer in B-cells inhibitor, alpha; IRF9: interferon regulatory factor 9. Almost 100% of the gene copies came from a single population. It is P. Pe´rez et al. / Journal of Autoimmunity 33 (2009) 99–108 107 noteworthy that 1p34.2 (D1S3721), 3q13.31-q13.32 (D3S2460), target organs of primary Sjogren’s syndrome. Proc Natl Acad Sci U S A 7p22.2-p22.1 (D7S1819), 10p13 (D10S1430), and 14q12 (D14S608) 2006;103:2770–5. [13] Daniels TE. Labial salivary gland biopsy in Sjogren’s syndrome. Assessment as loci harbor genes differentially expressed in SS LSG-epithelial cells a diagnostic criterion in 362 suspected cases. Arthritis Rheum 1984;27:147–56. (Table 4), and thus worthy of further fine mapping. Besides LAPTM5, [14] Vitali C, Bombardieri S, Jonsson R, Moutsopoulos HM, Alexander EL, LTF, IL6, CD44 and BCL2L2 genes or their respective proteins which Carsons SE, et al. Classification criteria for Sjogren’s syndrome: a revised version of the European criteria proposed by the American-European have been previously examined in SS, ZC3H12A, NASP and BCL2L2 Consensus Group. Ann Rheum Dis 2002;61:554–8. are new candidate genes for the disease. ZC3H12A is an MCP1 [15] Bravo ML, Valenzuela CY, Arcos-Burgos OM. Polymorphisms and phyletic (CCL2; MIM 158105)-induced protein that acts as a transcriptional relationships of the Paisa community from Antioquia (Colombia). Gene Geogr 1996;10:11–7. activator [53]. NASP interacts with histone H1 and that this inter- [16] Arcos-Burgos M, Muenke M. Genetics of population isolates. Clin Genet action occurs with high affinity, and also forms distinct, high 2002;61:233–47. specificity and functional complexes with histones H3 and H4 [54]. [17] Herrero J, Vaquerizas JM, Al-Shahrour F, Conde L, Mateos A, Diaz-Uriarte JS, et al. New challenges in gene expression data analysis and the extended BCL2L2, also known as BCL-W, suppress apoptosis induced by over GEPAS. Nucleic Acids Res 2004;32:W485–91. expression of either Bax or Bak [55]. Bak and its polymorphic gene [18] Herrero J, Al-Shahrour F, Diaz-Uriarte R, Mateos A, Vaquerizas JM, Santoyo J, have been shown to participate into the SS pathogenesis [56]. et al. GEPAS: a web-based resource for microarray gene expression data Taken together, our results show that both pro-apoptotic/anti- analysis. Nucleic Acids Res 2003;31:3461–7. [19] Pritchard JK, Stephens M, Donnelly P. Inference of population structure using apoptotic and IFN signaling inhibition/stimulation balances occur multilocus genotype data. Genetics 2000;155:945–59. in SS LSG-epithelial cells. Loci harboring the genes involved in these [20] Pritchard JK, Stephens M, Rosenberg NA, Donnelly P. Association mapping in pathways are candidates for disease susceptibility. structured populations. Am J Hum Genet 2000;67:170–81. [21] Kawasaki S, Kawamoto S, Yokoi N, Connon C, Minesaki Y, Kinoshita S, et al. Up- regulated gene expression in the conjunctival epithelium of patients with Sjogren’s syndrome. Exp Eye Res 2003;77:17–26. Acknowledgments [22] Ogawa N, Ping L, Zhenjun L, Takada Y, Sugai S. Involvement of the interferon- gamma-induced T cell-attracting chemokines, interferon-gamma-inducible We thank all the patients who participated in this study. This 10-kd protein (CXCL10) and monokine induced by interferon-gamma (CXCL9), in the salivary gland lesions of patients with Sjogren’s syndrome. Arthritis work was supported by Fondecyt (1080006, 1050192, and Rheum 2002;46:2730–41. 1020755), Chile (MJG, SA, CM), MECESUP-Postgrado and Beca [23] Eliasson L, Almstahl A, Lingstrom P, Wikstrom M, Carlen A. Minor gland saliva Conicyt (PP), Colciencias (2213-04-16715), Marshfield Foundation, flow rate and proteins in subjects with hyposalivation due to Sjogren’s and Rosario University (J-MA). syndrome and radiation therapy. Arch Oral Biol 2005;50:293–9. [24] Haghighat N, al-Hashimi I. The status of lactoferrin and total iron binding capacity of human parotid saliva in Sjogren’s syndrome. Clin Exp Rheumatol 2003;21:485–8. Appendix. Supplementary material [25] Hansen A, Reiter K, Ziprian T, Jacobi A, Hoffmann A, Gosemann M, et al. Dysregulation of chemokine receptor expression and function by B cells of Supplementary data associated with this article can be found in patients with primary Sjogren’s syndrome. Arthritis Rheum 2005;52:2109–19. [26] Ohyama Y, Nakamura S, Matsuzaki G, Shinohara M, Hiroki A, Oka M, et al. T- the online version at doi:10.1016/j.jaut.2009.05.001 cell receptor V alpha and V beta gene use by infiltrating T cells in labial glands of patients with Sjogren’s syndrome. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1995;79:730–7. References [27] Nakken B, Jonsson R, Brokstad KA, Omholt K, Nerland AH, Haga HJ, et al. Associations of MHC class II alleles in Norwegian primary Sjogren’s syndrome [1] Anaya J-M, Vega P, Correa PA, Kwon YJ, Brito M, Alliende C, et al. Sı´ndrome de patients: implications for development of autoantibodies to the Ro52 auto- Sjo¨gren. In: Anaya JM, Shoenfeld Y, Correa PA, Garcı´a-Carrasco M, Cervera R, antigen. Scand J Immunol 2001;54:428–33. editors. Autoinmunidad y Enfermedad Autoinmune. Medellin: Corporacio´ n [28] Manganelli P, Fietta P. Apoptosis and Sjogren syndrome. Semin Arthritis para Investigaciones Biolo´ gicas; 2005. p. 295–315. Rheum 2003;33:49–65. [2] Humphreys-Beher MG, Peck AB, Dang H, Talal N. The role of apoptosis in the [29] Loiseau P, Lepage V, Djelal F, Busson M, Tamouza R, Raffoux C, et al. HLA class I initiation of theautoimmune response in Sjogren’s syndrome. Clin Exp and class II are both associated with the genetic predisposition to primary Immunol 1999;116:383–7. Sjogren syndrome. Hum Immunol 2001;62:725–31. [3] Mitsias DI, Kapsogeorgou EK, Moutsopoulos HM. The role of epithelial cells in [30] Azuma T, Takei M, Yoshikawa T, Nagasugi Y, Kato M, Otsuka M, et al. Identi- the initiation and perpetuation of autoimmune lesions: lessons from Sjogren’s fication of candidate genes for Sjogren’s syndrome using MRL/lpr mouse syndrome (autoimmune epithelitis). Lupus 2006;15:255–61. model of Sjogren’s syndrome and cDNA microarray analysis. Immunol Lett [4] Perez P, Goicovich E, Alliende C, Aguilera S, Leyton C, Molina C, et al. Differ- 2002;81:171–6. ential expression of matrix metalloproteinases in labial salivary glands of [31] Fei HM, Kang H, Scharf S, Erlich H, Peebles C, Fox R. Specific HLA-DQA and patients with primary Sjogren’s syndrome. Arthritis Rheum 2000;43:2807–17. HLA-DRB1 alleles confer susceptibility to Sjogren’s syndrome and autoanti- [5] Perez P, Kwon YJ, Alliende C, Leyton L, Aguilera S, Molina C, et al. Increased body production. J Clin Lab Anal 1991;5:382–91. acinar damage of salivary glands of patients with Sjogren’s syndrome is par- [32] Kay RA, Hutchings CJ, Ollier WE. A subset of Sjogren’s syndrome associates alleled by simultaneous imbalance of matrix metalloproteinase 3/tissue with the TCRBV13S2 locus but not the TCRBV2S1 locus. Hum Immunol inhibitor of metalloproteinases 1 and matrix metalloproteinase 9/tissue 1995;42:328–30. inhibitor of metalloproteinases 1 ratios. Arthritis Rheum 2005;52:2751–60. [33] Emamian ES, Leon JM, Lessard CJ, Grandits M, Baechler MC, Gaffney PM, et al. [6] Goicovich E, Molina C, Perez P, Aguilera S, Fernandez J, Olea N, et al. Enhanced Peripheral blood gene expression profiling in Sjo¨gren’s syndrome. Genes degradation of proteins of the basal lamina and stroma by matrix metal- Immun 2009 Apr 30 [Epub ahead of print]. loproteinases from the salivary glands of Sjogren’s syndrome patients: [34] Wildenberg ME, van Helden-Meeuwsen CG, van de Merwe JP, Drexhage HA, correlation with reduced structural integrity of acini and ducts. Arthritis Versnel MA. Systemic increase in type I interferon activity in Sjo¨gren’s Rheum 2003;48:2573–84. syndrome: a putative role for plasmacytoid dendritic cells. Eur J Immunol [7] Anaya JM, Tobon GJ, Vega P, Castiblanco J. Autoimmune disease aggregation in 2008;38:2024–33. families with primary Sjogren’s syndrome. J Rheumatol 2006;33:2227–34. [35] Molina C, Alliende C, Aguilera S, Kwon YJ, Leyton L, Martinez B, et al. Basal [8] Foster H, Stephenson A, Walker D, Cavanagh G, Kelly C, Griffiths I. Linkage lamina disorganisation of the acini and ducts of labial salivary glands from studies of HLA and primary Sjogren’s syndrome in multicase families. Arthritis patients with Sjogren’s syndrome: association with mononuclear cell infil- Rheum 1993;36:473–84. tration. Ann Rheum Dis 2006;65:178–83. [9] Anaya JM, Delgado-Vega AM, Castiblanco J. Genetic basis of Sjogren’s [36] Philchenkov A. Caspases: potential targets for regulating cell death. J Cell Mol syndrome. How strong is the evidence? Clin Dev Immunol 2006;13:209–22. Med 2004;8:432–44. [10] Killedar SJ, Eckenrode SE, McIndoe RA, She JX, Nguyen CQ, Peck AB, et al. Early [37] Rothwarf DM, Karin M. The NF-kappa B activation pathway: a paradigm in pathogenic events associated with Sjogren’s syndrome (SjS)-like disease of the information transfer from membrane to nucleus. Sci STKE 1999. 1999:RE1. NOD mouse using microarray analysis. Lab Invest 2006;86:1243–60. [38] Fay MJ, Longo KA, Karathanasis GA, Shope DM, Mandernach CJ, Leong JR, et al. [11] Hjelmervik TO, Petersen K, Jonassen I, Jonsson R, Bolstad AI. Gene expression Analysis of CUL-5 expression in breast epithelial cells, breast cancer cell lines, profiling of minor salivary glands clearly distinguishes primary Sjogren’s normal tissues and tumor tissues. Mol Cancer 2003;2:40. syndrome patients from healthy control subjects. Arthritis Rheum [39] Martin AG, San-Antonio B, Fresno M. Regulation of nuclear factor kappa B 2005;52:1534–44. transactivation. Implication of phosphatidylinositol 3-kinase and protein [12] Gottenberg JE, Cagnard N, Lucchesi C, Letourneur F, Mistou S, Lazure T, et al. kinase C zeta in c-Rel activation by tumor necrosis factor alpha. J Biol Chem Activation of IFN pathways and plasmacytoid dendritic cell recruitment in 2001;276:15840–9. 108 P. Pe´rez et al. / Journal of Autoimmunity 33 (2009) 99–108

[40] De Laurenzi V, Melino G. Apoptosis. The little devil of death. Nature [49] Zhang L, Pagano JS. Structure and function of IRF-7. J Interferon Cytokine Res 2000;406:135–6. 2002;22:95–101. [41] Strasser A. The role of BH3-only proteins in the immune system. Nat Rev [50] Bave U, Nordmark G, Lovgren T, Ronnelid J, Cajander S, Eloranta ML, et al. Immunol 2005;5:189–200. Activation of the type I interferon system in primary Sjogren’s syndrome: [42] Sisto M, Lisi S, Castellana D, Scagliusi P, D’Amore M, Caprio S, et al. Autoan- a possible etiopathogenic mechanism. Arthritis Rheum 2005;52: tibodies from Sjogren’s syndrome induce activation of both the intrinsic and 1185–95. extrinsic apoptotic pathways in human salivary gland cell line A-253. [51] Vanbervliet B, Bendriss-Vermare N, Massacrier C, Homey B, de Bouteiller O, J Autoimmun 2006;27:38–49. Briere F, et al. The inducible CXCR3 ligands control plasmacytoid dendritic cell [43] Lauricella M, Emanuele S, D’Anneo A, Calvaruso G, Vassallo B, Carlisi D, et al. responsiveness to the constitutive chemokine stromal cell-derived factor 1 JNK and AP-1 mediate apoptosis induced by bortezomib in HepG2 cells via (SDF-1)/CXCL12. J Exp Med 2003;198:823–30. FasL/caspase-8 and mitochondria-dependent pathways. Apoptosis [52] Arumugam TV, Okun E, Tang SC, Thundyil J, Taylor SM, Woodruff TM. Toll- 2006;11:607–25. Like receptors in Ischemia-Reperfusion Injury. Shock 2008 Nov 11 [Epub [44] Kwon YJ, Perez P, Aguilera S, Molina C, Leyton L, Alliende C, et al. Involvement ahead of print]. of specific laminins and nidogens in the active remodeling of the basal lamina [53] Liang J, Wang J, Azfer A, Song W, Tromp G, Kolattukudy PE, et al. A novel of labial salivary glands from patients with Sjogren’s syndrome. Arthritis CCCH-zinc finger protein family regulates proinflammatory activation of Rheum 2006;54:3465–75. macrophages. J Biol Chem 2008;283:6337–46. [45] Alberts B, Johnson A, Lewis J, Raff M, Roberts K, Walter E. Internal Organization [54] Wang H, Walsh ST, Parthun MR. Expanded binding specificity of the human of the cell. In: GS T, editor. Molecular biology of the cell. Taylor & Francis histone chaperone NASP. Nucleic Acids Res 2008;36:5763–72. Group; 2007. [55] Zhai D, Jin C, Huang Z, Satterthwait AC, Reed JC. Differential regulation of Bax [46] Oliveira SS, Morgado-Diaz JA. Claudins: multifunctional players in epithelial and Bak by anti-apoptotic Bcl-2 family proteins Bcl-B and Mcl-1. J Biol Chem tight junctions and their role in cancer. Cell Mol Life Sci 2007;64:17–28. 2008;283:9580–6. [47] Fievet B, Louvard D, Arpin M. ERM proteins in epithelial cell organization and [56] Delgado-Vega AM, Castiblanco J, Go´ mez LM, Diaz-Gallo LM, Rojas-Villarraga A, functions. Biochim Biophys Acta 2007;1773:653–60. Anaya JM. BCL2 antagonist killer 1 (BAK1) polymorphisms influence the risk of [48] Takaoka A, Yanai H. Interferon signalling network in innate defence. Cell developing autoimmune rheumatic diseases in women. Ann Rheum Dis 2009 Microbiol 2006;8:907–22. Mar 11 [Epub ahead of print].